Target-Decoy-Based False Discovery Rate Estimation for Large-Scale Metabolite Identification

• DOI: 10.1021/acs.jproteome.8b00019
• Publication Date: 2018-05-23T13:49:15Z
• Authors: Xusheng Wang; Drew R. Jones; Timothy I. Shaw; Ji-Hoon Cho; Yuanyuan Wang; Haiyan Tan; Boer Xie; Suiping Zhou; Yuxin Li; Junmin Peng
• Keywords: 0301 basic medicine; Models, Theoretical; High-Throughput Screening Assays; Small Molecule Libraries; 03 medical and health sciences; Databases as Topic; Tandem Mass Spectrometry; Yeasts; Metabolome; Metabolomics; False Positive Reactions; Algorithms; Software

Metabolite identification is a crucial step in mass spectrometry (MS)-based metabolomics. However, it is still challenging to assess the confidence of assigned metabolites. We report a novel method for estimating the false discovery rate (FDR) of metabolite assignment with a target-decoy strategy, in which the decoys are generated through violating the octet rule of chemistry by adding small odd numbers of hydrogen atoms. The target-decoy strategy was integrated into JUMPm, an automated metabolite identification pipeline for large-scale MS analysis and was also evaluated with two other metabolomics tools, mzMatch and MZmine 2. The reliability of FDR calculation was examined by false data sets, which were simulated by altering MS1 or MS2 spectra. Finally, we used the JUMPm pipeline coupled to the target-decoy strategy to process unlabeled and stable-isotope-labeled metabolomic data sets. The results demonstrate that the target-decoy strategy is a simple and effective method for evaluating the confidence of high-throughput metabolite identification.